Predict SETI@Home Rank, Percentile and Medallion Accomplishment

[ Compute My Personal Milestones Predictions ]

[ Compute When I Overtake Another ]

[ Percentile Table | Rank Table | Golden WU Table ]

[ Data to Paste into Spreadsheet ]

[ Table Column Descriptions | About The Tables | Download Source ]

[ Contact Author | Here is my email address ... Include my SETI Data in Tables  ]

Thanks for the memories... And thanks for the kindnesses that many SETI Crunchers have sent my way. Your help and good thoughts have made my four+ years with SETI personally (though certainly not financially) rewarding.

The announced 12/15/2005 closure of the SETI/Classic servers casts doubt upon the validity of any goal projections much beyond that date.

This will be the last publication for use in projection. I may, if the data is still available, run a final retrieval after the data servers shut down and publish tables which tell the "final" relationships between WU's, Rank, and Percent.

I do not plan to migrate this project to SETI/BOINC. For anyone who wishes to do so, I will be pleased to facilitate your understanding. Contact me via email.

Percent, 12/10/2005 2:56:19 AM UTC

Rank, 12/10/2005 2:56:19 AM UTC

Golden Work Unit, 12/10/2005 2:56:19 AM UTC

Chart Detail: Chart detail has been moved to a separate file.

To use the "Goal" computation function, enter the three items about yourself, and click "Compute". If your WU total is not "as of right now", tell how many days (and/or hours) ago it had that value. Decimal numbers are okay. Look in the tables just below for the results. The projection will be blank if your rate is too low to reach that level. A notes column explains reasons for blank projections and gives accuracy cautions.

Your computer clock is set to
If that is not the correct date and time, the computed date results will not be correct!

To use the "Catch 'em" computation function, enter the data about yourself and the data about the participant that you are chasing, and click "Compute". If the WU totals are not "as of right now", tell how many days and/or hours ago it had that value. Decimal numbers are okay. Look in the tables just below for the results. The projection will be blank if your rate is too low to catch him.

Your computer clock is set to
If that is not the correct date and time, the computed date results will not be correct!

Table Column Descriptions

Pct or Rank		Numbers in this column are titles for their data lines. The data in the line describes this particular Percentile or Rank
(Rank) or (Pct)		Numbers in this column are Ranks or Percents that correspond to the line title.
WU Now		This is the number of WU projected to the time that the pages is displayed.
WU/.001%		This is the number of WU spanned by .001% at this level.
WU/Day		This number is the "slope" or "rate" in the regression formula derived from the observed WU totals. It tells how much, on average, this Percentile or Rank changes each day. The Observations can be extrapolated to future days as follows: WU Needed in "N" days = I-cept + N * WU/Day The larger the value of N, the less accurate the extrapolation is likely to be. Linearity assumptions are less valid over longer periods. WU/Day values should be monotonic non-decreasing as one moved down the chart. However, the numbers don't always cooperate. The statistics reflect local, probably transient variations in the "big picture." They will fluctuate over time.
I-cept		This number is the Y-Intercept in the regression formula derived from the observed WU totals. It tells the Current Number of WU needed, on average, to achieve this Percentile or Rank.
R2		R-Squared tells the fraction of variance in Number of WU's (the dependent variable) that is attributed to passage of days (the independent variable).
Observed WU		Numbers in these columns tell the Number of WU needed to achieve the particular Percentile or Rank on that day. Only the last 5 observations are shown in the table. Older data is included in the computations. The numbers are based on individual statistics maintained by SETI@home. When exact samples are not available (which is most of the time), the Number is derived by linear interpolation between surrounding samples. Percentiles values are recomputed to 6 decimal places to improve results at high end of percentile table. Derived values are rounded to two decimal places, which helps characterize lower percentile or rank results more clearly. Column titles tell the age of the data relative to the time stamp of the last retrieval session, usually Saturday night/ Sunday morning.
Projected WU		These results are derived by applying the regression formula to project out 1, 7 and 30 days. Linearity is assumed. Projection is based on Least Squares Regression computed for each Pct and Rank level. Observations are weighted inversly proportional to age, e.g. w(today) = 1; w(2 days ago) = 1/3.
Interpolation Span WU		This is the number of WU separating the observations between which the "Observed WU" result is interpolated. It is a measure of the density of the sampling and of the sparcity of occupants in that part of SETI@home's tables.
Interpolation Span Pct or Rank		This is the difference between the percentile or rank values between which the "Observed WU" result is interpolated.

Tables show Work Unit (WU) for various Percents and Ranks for participants in SETI@home.

Lines with R-Squared less than 90% are listed in smaller font.

Observed data is extracted from SETI@home User Stats.

Additional email addresses are needed. But, please, only for participants who have completed at least 2500 WU and whose address is not in their public profile. I will not share them and I will not use them to send emails ... just to extract SETI data for these charts.

WU counts are rounded to two decimal places. In most cases, WU counts are inferred by linear interpolation between actual observations.

Percentiles are computed to 6 decimal digits to improve results for percentile scores above 99%.

The entire set of observations are listed at the bottom, suitable for en masse copy and paste into a spreadsheet. You are welcome to analyze to your heart's content. Let me know if you use these.

The columns under the title "Interpolation Span" were added to help assess the need for additional addresses to sample. The WU column shows the difference in submitted Work Units for the samples just above and just below the target. Same idea for Pct and Rank.

4/13/03 - Goal Attainment computation added.

4/18/03 - Percentile computation correction.

4/25/03 - Dropping Percentile 99.995 and Ranks 100 and 200 until email addresses in those ranges are available.

4/29/03 - Adding Golden Work Unit (GWU) display. Golden WU is the point at which WU Count equals Rank.

5/8/03 - Add offsets and Dates in Compute Milestones function.

5/16/03 - Add option to produce projections in popup window.

5/17/03 - Minor bug fix for low WU totals.

5/18/03 - Add Medallion Goals!

5/19/03 - Change offset to days and hours.

5/20/03 - Report if Golden WU within 100 day horizon.

5/21/03 - Add "WU Now" column showing number of Work Units "as of the time the page is loaded by the user."

5/24/03 - Add "Overtake another participant" projection.

6/1/03 - Highlight key table columns.

6/3/03 - Refresh display every 15 minutes to keep "WU Now" column up to date.

7/5/03 - Extend scope of printable popup to 180days + 1 item.

7/20/03 - Eliminate separate button for "on-screen only" results display.

7/31/03 - Split out data for copy/paste.

3/26/04 - Improve consistency in projected date display. (Esp. Opera browser.)

10/04/04 - Add chart of key percent rates of movement biweekly since start of 04.